Tengda Han

About

Tengda Han is a computer vision researcher whose work centers on human motion understanding, action forecasting, and intelligent human-robot interaction. His most influential contribution, "Human Pose Forecasting via Deep Markov Models" (2017), has garnered 43 citations and represents a significant advance in predicting long-range human body motion from 3D skeleton sequences — a capability with far-reaching implications for autonomous driving, visual surveillance, and robotic systems. By applying Deep Markov Models to this problem, Han pushed beyond the limitations of short-term forecasting that had constrained earlier approaches, opening new possibilities for temporally extended prediction. Complementing this work, Han's research on "Human Action Forecasting by Learning Task Grammars" (2017) addresses the practical challenge of anticipating human intentions during complex, repetitive real-world tasks — a critical capability for robotic assistants operating alongside people. Together, these contributions establish Han as a thoughtful contributor to the intersection of probabilistic modeling and human motion analysis. His research tackles problems where machine perception must anticipate rather than merely react, laying groundwork for safer, more responsive AI systems in dynamic environments shared with humans.

Research Focus

Key Achievements

Top Papers

1. 1
   Human Pose Forecasting via Deep Markov Models

   43 citations · 2017
2. 2
   Human Action Forecasting by Learning Task Grammars

   12 citations · 2017
3. 3
   Human Pose Forecasting via Deep Markov Models

   2 citations · 2017

Key Collaborators

Contact & Links